Model Uncertainty Assessment for the Anchorage Length Formulation of fib Model Code 2020

In the design of pre-tensioned concrete elements, it is important to avoid the anchorage failure of tendons. This can be done by predicting the anchorage length through any of the empirical or semiempirical models provided by the main design codes. However, the models in these codes are based on experimental evidence gathered many years ago, when the 7-wire strands manufacturing process was significantly different from now. The objective of this contribution is to evaluate the model uncertainty for the anchorage length formulation of the last draft of the fib Model Code 2020 and to ensure that all the main influencing factors are properly considered. Since a direct measure of anchorage length is not experimentally feasible, a large data set of flexural tests has been gathered from the scientific literature, and each failure mode has been compared to the corresponding anchorage length prediction, to obtain a censored sample of the model uncertainty from which an appropriate probability distribution could be estimated. Then, the minimum values of the partial safety factors that consider both model and material uncertainties and ensure the achievement of the target reliability in anchorage length predictions have been derived through Monte Carlo Simulations. Finally, these results have been validated through a FORM approach for some case studies and compared to the current fib Model Code 2020 proposed partial safety factors, showing that they guarantee the achievement of the expected level of reliability in all the evaluated situations.